I remember reading about speaking in an encouraging manner to agentic AI leading to better results, but I can’t seem to find a citation for this.